This is the mail archive of the gsl-discuss@sources.redhat.com mailing list for the GSL project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Bug: gamma_inc



Hi!

I have now patched the incomplete gamma function gamma_inc.c, using
the modified version of Lentz' algorithm as given in Numerical
Recipes.  

I have tested the results against Mathematica, using only such a,x
pairs that the continued fraction expression was used.  a and x were
chosen randomly.

Range of a     max abs difference      CPU time
20-     100       1.6e-14               ~45s

20-   10000       1.1e-11               ~90s

20- 1000000       5  e-04              ~270s

For each of the three ranges, I generated 1 million a,x pairs; all
computations were done on a PIII@866MHz.  Difference is against
numerical evaluation with Mathematica 4.0.  For the range of up to
1e6, most errors were much smaller than the max error, typically
less than 1e-10.

A diff-file for patching gamma_inc.c against the version in gsl-1.0 is
included.

Hans

Attachment: gamma_inc.c.diff
Description: Binary data



---------------------------------------------------------------------
Dr. Hans Ekkehard Plesser             Tel.  :           +47 6494 8832
Physics Section / ITF                 Fax   :           +47 6494 8810
Agricultural University of Norway     e-mail: hans.plesser@itf.nlh.no
N-1432 Ås, Norway		      WWW   :    arken.nlh.no/~itfhep
---------------------------------------------------------------------

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]